Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems Power supply upgrade

Oh boy, this is exciting - I am really looking forward to hearing what the results will show for the C61.

Good work !

While it�s fresh in my mind, I got the C61R working with a linear power supply at the weekend, and eagerly hooked it up alongside the 861V4, using just the analogue outputs, and no DSP modes. I didn�t really expect to upset the apple-cart, but I was keen to see how much an LPS would improve another controller. I used a Power One supply with the high current output set to 5.0V, rather than 5.2V, and no current sensing.

It occurred to me what a fascinating exercise this was, as no-one was likely to have heard before what a G-series controller could do, unshackled by the SMPS (which turned out to be a lot). My impression was the C61 benefitted from the linear supply at least as much as the 861, and perhaps more. Having gotten used to the modified 861, I hadn�t been too impressed with the C61 beforehand. With the linear supply, it was greatly improved: big, bold, dynamic and transparent, without the slight muddiness and harshness of the stock unit.

Comparing the 861 & C61 was surprisingly difficult, and I spent longer than I expected trying to choose. Many times I felt like I was listening to the same amplifier, and might have been happy with either. After many (happy) hours it finally became clear that the 861 was more neutral and refined, while the C61 was punchy and dynamic. This was nothing like comparing the stock G68 and stock 861, or the 861 with switching and linear supplies, where the differences were clear as daylight.

It was only when digging really deep that I finally heard a clear advantage to the 861, which had better low-level resolution, darker backgrounds, and more subtle phrasing and musical interplay. On the other hand, the C61 was good with the skeleton of the music, and could hold on to a rhythm better, but was a bit more two-dimensional. The fact (to my mind) that the simple addition of an LPS to a G series controller not only brings it up to the level of the 861, but nearly to the level of the 861/LPS, is quite remarkable. That makes the functionality, form factor and feature-set of the C61/G61/G68 rather attractive, but having heard the 861/LPS, it�s hard to step back and I�ll probably be keeping it. Which is what I wanted to know.

Nick

Nick, I don't want to re-read the whole thread again (yet!). Are you talking about analogue outputs only here?

Edit: Perhaps I should have properly read the post...

I figured someone would ask that, so I answered it in the first line. Yes, I have only been using analogue outputs so far. I don't have any DSP speakers, but hope to rectify that shortly. I can only speculate whether the linear PSU improves digital outputs, but I'd be surprised if there wasn't some benefit.

Nick

Very interesting Nick, I'm not surprised by your findings. I'd be most interested to hear the opinions of a blind listening lay (ie non hifi enthusiast) person, a partner for example - I suspect the differences between the two units is actually quite small...

Meridian are faced with a whole host of euro nanny regulations and cost issues when they're designing a product and this can lead to compromises in potential audio quality.

Higher quality power supplies can be built and they will achieve a higher audio quality but at much higher cost as a new PSU design need new certifications etc (8k Ruby anniversary implementation)

The mains input filter on the G series for example is of low quality and includes the inevitable fuse - for audio quality considerations only best bypassed all together IMHO, as are any fuses in the mains. I use euro schuko plugs and sockets for this very reason.

For circuitry containing only digital signals a bank of batteries in conjunction with a Jung type superreg circuit is very effective indeed.

However, batteries are a disaster on any analogue audio boards due to their high impedance which softens the dynamics. The Sooloos Control 15 contains a digital only output board inside its base...

From the internal pics posted of the MS600 there would appear to be considerable spare real estate inside the unit for a linear supply conversion.

Avondale Audio in the UK come highly recommended for linear PSU's, Les is aloways open to custom projects and has many many years experience. Googled searches for Squeezebox PSU upgrades will reveal others.

IMHO there are two quality bottlenecks present in M components, the first is the use of SMPS in conjunction with cheap off the shelf input filtration modules. The second is the SPDIF (inc speakerlink) interface.

I couldn't agree more.

Dpaws, why are you saying SL would be a bottleneck? It's differential is it not thus somwhat immune to interference and with good signal integrity?

Traditional thinking suggests that digital circuitry needs very clean (regenerated?) and highly regulated power supplies. Impedance here is not so critical as there are none of the power surges found with an analogue signal, other than being able to go from 0 to 1 very quickly to achieve the squarest front edge to the digital signal. Digital circuitry is particularly vulnerable to EMI/EFI interference and ideally should be shielded.

The largest gains from a low impedance linear PSU are with analogue circuitry. Typically overspec the power supply 4:1 and utilised the lowest ESR components. Transformer design is critical, never skimp in this area.

Sorry Crion, first I must state that I am no expert - I'm sure others can elaborate in a far more technically coherent manner. There is some insight here : http://www.wolfsonmicro.com/documen...spdif_digital_interface_transceivers.pdf and here : http://www.scalatech.co.uk/papers/aes93.pdf

The SL is a psudo balanced SPDIF connection, not a true differential digital interface. However, its not the influence of the signal passing through the cable itself that I am referring to, it is the jitter incurred by the interface conversion process itself.

For analogue power amp users the 808.3 with internal ID40 is the optimum configuration in this respect. On the DSP side you would want an 8k with internal ID40 card, although how the two cards (one required in each speaker) might synchronise with each other is beyond my technical knowledge...

If I understand speakerlink correctly, its like differential SPDIF on an ethernet cable. Which means that the audio clock and audio data share the same channel. That inevitably means that the audio clock is compromised, as its modulated by the data itself.

Perhaps ethernet is better than coax in this respect, but I wouldn't expect so. The audio clock, whether its bit clock or word clock, should be treated as a valuable and fragile analogue signal, and be very well protected with its own connection.

Introducing a new, proprietary, digital audio connection should have been a golden opportunity for Meridian to do it right, but it seems to me that they have missed it. I still have a lot of confidence in what Meridian do, but I'd sure like to know what their reasoning was for introducing speakerlink (or MMHR for that matter).

The Wolfson paper isn't useful, but the Scalatech paper is essential reading, as is this: http://www.tnt-audio.com/clinica/diginterf1_e.html. Note the bibliography, which includes lots of work by Dunn, Dunn & Hawksford.

Notice how both papers clearly describe better than I can the ideal digital audio replay architecture. Its figure 39 in the Dunn & Hawksford paper and the Backwards Clock Configuration in part 1.2 of the TNT paper. In each case, they describe a DAC-Master configuration, where the master audio clock is located with the DAC itself. This is exactly where its needed, with no processing or connections between the clock and the DAC to currupt the timing information. This contrasts with the usual Transport Master configuration, where the master clock is in the transport, and has to travel forwards to the DAC. In most systems, that have SPDIF, MMHR or speakerlink connections, the clock is additionally degraded by data modulation, as described so well in the Scalatech paper.

Nick

I'm guessing that Meridian is confident that up-sampled audio (lower inherent jitter) + MHR encryption (minimizes correlated jitter in the xfer) + FIFO in the speaker is sufficient to lower jitter to inaudible levels in the DAC.

I'm also guessing that Meridian didn't want to introduce clocking which was linked between components and/or played it safe with their known technology.

Bear in mind that SpeakerLink was simply to meet a requirement from the custom install market to be able to utilize in-wall RJ45 cabling. Meridian didn't expect it to sound better than SPDIF - they put down any improvement to the improved encoding and decoding circuitry at either end of the link.

A true differential digital output has additional "white noise" imposed over it which is removed at the other end - apologies, I can't find the appropriate link where I read about this.

The debate started in the 80's when CD players became two box units. The advantages gained from the reduced PSU/EMI interference were countered by the jitter incurred between the two units via SPDIF.

The latest thinking is the one box transport / DAC but with an external PSU.

With regard to SPDIF alternatives, DPA played with non standard fibre optic connections many years ago, reborn recently by Weiss between the Jason transport and Medea DAC in their reference products.

PS Audio use an i2s feed over an HDMI cable. I am sure there are many other examples but none have been adopted universally.

On the positive side, M finally realised that similar length SL runs to individual speakers was technically superior to running a cable to one speaker and then daisy chaining through additional interfaces.